A major obstacle to tobacco control in non-Western countries is the lack of reliable local evidence about the extent to which tobacco is currently causing death from particular diseases, and about any large changes in these hazards. The most efficient, direct means of getting this information is through prospective studies of large populations in which baseline information on tobacco use and a few other important characteristics is collected, and then, in the following years, deaths and their causes are monitored. We have already established large collaborative, prospective studies in nine distinct populations in six non-Western countries to assess the current relevance of tobacco (and a few other major causes of death) to cause-specific mortality, and to monitor the evolution over the next few decades of the relevance of tobacco and other risk factors to cause-specific mortality. This application relates to support for six of these studies in four countries, which together will include two million adults. The following items are recorded for each participant: tobacco and alcohol habits (both by locally appropriate questions), blood pressure, weight and height, peak expiratory airflow, personal identifiers (with consent to long-term follow-up), previous history of selected diseases, and a few questions relevant to each population. Follow-up requires monitoring, by a variety of means, the underlying cause of all deaths in the cohorts over the coming years. In addition, a few thousand in each population will be resurveyed to monitor changes in characteristics and to allow appropriate correction for the "regression dilution bias." Early follow-up from the first study to be established, of 250,000 men in China, has already been carried out, so the feasibility of this approach is known (although adaptations for each particular population will be needed). Follow-up in all six populations will require funds both to carry out the fieldwork and analysis and to provide training opportunities to our collaborators in both the generation and the politically effective use of reliable epidemiologic evidence. The information gained through these studies will feed back into public consciousness, public health and future research strategies. We already have evidence that this has taken place in China, where the earliest study was established.